HAL Id: hal-03157431
https://hal.sorbonne-universite.fr/hal-03157431
Submitted on 3 Mar 2021
HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
Multi-model Meteorological and Aeolian Predictions for Mars 2020 and the Jezero Crater Region
C. Newman, M de la Torre Juárez, J. Pla-García, R. Wilson, S. Lewis, L.
Neary, M. Kahre, F. Forget, A. Spiga, M. Richardson, et al.
To cite this version:
C. Newman, M de la Torre Juárez, J. Pla-García, R. Wilson, S. Lewis, et al.. Multi-model Meteoro- logical and Aeolian Predictions for Mars 2020 and the Jezero Crater Region. Space Science Reviews, Springer Verlag, 2021, 217 (1), pp.20. �10.1007/s11214-020-00788-2�. �hal-03157431�
https://doi.org/10.1007/s11214-020-00788-2
Multi-model Meteorological and Aeolian Predictions for Mars 2020 and the Jezero Crater Region
C.E. Newman1·M. de la Torre Juárez2·J. Pla-García3,4·R.J. Wilson5·S.R. Lewis6· L. Neary7·M.A. Kahre5·F. Forget8·A. Spiga8,9·M.I. Richardson1·F. Daerden7· T. Bertrand10,5·D. Viúdez-Moreiras3·R. Sullivan11·A. Sánchez-Lavega12· B. Chide13·J.A. Rodriguez-Manfredi3
Received: 20 May 2020 / Accepted: 26 December 2020 / Published online: 8 February 2021
© The Author(s) 2021
Abstract Nine simulations are used to predict the meteorology and aeolian activity of the Mars 2020 landing site region. Predicted seasonal variations of pressure and surface and atmospheric temperature generally agree. Minimum and maximum pressure is predicted at Ls∼145◦and 250◦, respectively. Maximum and minimum surface and atmospheric tem- perature are predicted at Ls∼180◦and 270◦, respectively; i.e., are warmest at northern fall
The Mars 2020 Mission
Edited by Kenneth A. Farley, Kenneth H. Williford and Kathryn M. Stack
Supplementary Information The online version contains supplementary material available at https://doi.org/10.1007/s11214-020-00788-2.
B
C.E. Newmanclaire@aeolisresearch.com 1 Aeolis Research, Tucson, AZ, USA
2 Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA 91001, USA 3 Centro de Astrobiología (CSIC-INTA), 28850 Madrid, Spain
4 Space Science Institute, Boulder, CO 80301, USA 5 Ames Research Center, Mountain View, CA, USA 6 The Open University, Milton Keynes, UK
7 Belgian Institute for Space Aeronomy, Brussels, Belgium
8 Laboratoire de Météorologie Dynamique/Institut Pierre Simon Laplace (LMD/IPSL), Sorbonne Université, Centre National de la Recherche Scientifique (CNRS), École Polytechnique, École Normale Supérieure (ENS), 75005 Paris, France
9 Institut Universitaire de France, 75005 Paris, France
10 LESIA, Observatoire de Paris, Université PSL, CNRS, Sorbonne Université, Université de Paris, 92195, Meudon, France
11 Cornell Center for Astrophysics and Planetary Science, Cornell University, Ithaca, NY 14853, USA 12 Universidad del País Vasco (UPV/EHU), Bilbao, Spain
13 Institut Supérieur de l’Aéronautique et de l’Espace (ISAE), Toulouse, France
equinox not summer solstice. Daily pressure cycles vary more between simulations, possibly due to differences in atmospheric dust distributions. Jezero crater sits inside and close to the NW rim of the huge Isidis basin, whose daytime upslope (∼east-southeasterly) and night- time downslope (∼northwesterly) winds are predicted to dominate except around summer solstice, when the global circulation produces more southerly wind directions. Wind predic- tions vary hugely, with annual maximum speeds varying from 11 to 19 ms−1and daily mean wind speeds peaking in the first half of summer for most simulations but in the second half of the year for two. Most simulations predict net annual sand transport toward the WNW, which is generally consistent with aeolian observations, and peak sand fluxes in the first half of summer, with the weakest fluxes around winter solstice due to opposition between the global circulation and daytime upslope winds. However, one simulation predicts transport toward the NW, while another predicts fluxes peaking later and transport toward the WSW.
Vortex activity is predicted to peak in summer and dip around winter solstice, and to be greater than at InSight and much greater than in Gale crater.
Keywords Mars·Meteorology·Aeolian·Atmosphere·Dust devils·Mars 2020· Jezero crater
1 Introduction
The Mars 2020 Perseverance rover and accompanying Ingenuity Helicopter will land on Mars on 18 February 2021. On Mars, the areocentric solar longitude, Ls, will be∼5◦at this time, corresponding to local (northern hemisphere) spring of Mars Year (MY) 36. The cen- ter of the landing ellipse has longitude 77.43◦E, latitude 18.47◦N, and elevation−2.55 km.
This puts the landing site on the northwestern slopes of Jezero crater, a∼49 km diameter impact crater sitting on the northwestern slopes of the∼1500 km diameter Isidis basin, on the global-scale topographic dichotomy boundary. This provides an opportunity to measure both the near-surface atmospheric circulation and the aeolian activity due to that circula- tion at an interesting new location on the Martian surface, one at which the circulation is influenced by both regional (dichotomy boundary and Isidis basin) and local (Jezero crater) topography. Over the course of the mission the rover may drive a significant distance, allow- ing it to quantify the changing meteorology and circulation – and how those changes affect aeolian activity – across a wider region.
The set of instruments carried by Mars 2020, described in Sect.2and in other papers in this special issue, are able to monitor a wide range of variables that are relevant to mete- orological and aeolian studies. They will measure both meteorological variables (pressure, temperature, winds, and water vapor) and the forcing that drives their behavior over time (e.g. aerosol abundances and properties, and radiation fluxes at the surface). These measure- ments will enable investigations of processes operating on timescales from seconds to years, ranging from understanding the statistics of convective vortices and atmospheric turbulence to determining the impact of local topography or atmospheric dust opacity on near-surface wind stress, sand motion, and dust lifting. Images of the surface, combined with measure- ments of the local circulation, will provide insight into aeolian features such as ripples, dunes, ventifacts, and dusty convective vortex (dust devil) tracks. Changes detected on the surface or rover deck will be correlated with winds and vortex activity to shed light on ae- olian processes and the threshold conditions that must be exceeded for dust lifting or active saltation.
The model predictions and intercomparisons presented here are intended to provide con- text for such meteorological and aeolian studies, but also provide useful information for science teams and engineers working to refine mission operations. In addition, Objective D3 of the Mars 2020 mission is to make surface weather observations to validate global atmospheric models (Farley et al.2020, this journal). As noted in the 2020 Mars Explo- ration Program Analysis Group (MEPAG) Goals Document (Banfield2020): “Numerical modeling of the atmosphere is critical to understanding atmospheric and climate processes.
Models provide dimensional and temporal context to necessarily sparse and disparate ob- servational datasets [. . . ] and constitute a virtual laboratory for testing whether observed or inferred conditions are consistent with proposed processes.” The seven models used in this study (with output at two different horizontal resolutions presented for two of them, making nine simulations in total) represent a significant fraction of the Mars atmospheric models in current use. The results obtained during this intercomparison exercise should therefore enable a rapid assessment of how well current Mars models are able to predict the near- surface atmospheric state at a new location on Mars, and pave the way for more detailed investigations of model-data discrepancies and their causes.
The near-surface atmosphere is where many key processes occur that control the entire atmospheric circulation, from surface exchange of heat, momentum, and trace gases (in- cluding water vapor) to surface heating that drives daytime planetary boundary layer (PBL) convection (e.g. Stull1988). The greatest source of climate variability on Mars is associated with major dust storms, due to the large impact of dust loading on the thermal state of the thin Martian atmosphere (e.g. Gierasch and Goody1972; Wolff et al.2017). Yet we still do not understand what controls the onset, evolution, and decay of such storms, in particular the processes involved in dust lifting by wind stress, which likely drives onset of such storms (Newman et al.2002a,b; Kok2010; Newman and Richardson2015; Musiolik et al.2018;
Swann et al.2020). Our knowledge of near-surface meteorology is mostly gleaned from the few surface missions to date, which have largely provided bulk variables (temperature, pres- sure, horizontal wind, etc.) at a single height above the surface (e.g. Martínez et al.2017).
Unfortunately, real understanding of PBL and surface processes on Mars – such as the ver- tical mixing of heat and momentum and the exchange of both with the surface – require the measurement of variables at multiple heights and/or fluxes of heat, momentum, etc. (Rafkin et al.2009; Newman et al.2020). For Earth atmospheric models, parameterizations of these processes are based on numerous rigorous field observations. For Mars, parameterizations are typically adapted from those developed for Earth, with the choice of scheme and its parameters then determined by comparing their predictions of bulk variables with Mars ob- servations. Schemes have also been developed by using Mars Large Eddy Simulations (LES) to provide simulated PBL measurements (e.g. Temel et al.2021), but the problem remains of how to validate the LES results themselves. Overall, a wide range of schemes and param- eter values are used in Mars models, in large part due to this lack of information on what the correct choices are. Vertical grid spacings, especially near the surface where profiles of temperature and wind often change rapidly with height, also vary widely between models, as does the specific 4-D distribution of atmospheric dust, choice of roughness length and surface thermal properties, and even how variables are extrapolated from model layers to a given height above the surface (see Sect.3.8). As a result, and as shown in this study, mod- els typically differ widely in their predictions of the near-surface atmosphere. Differences in the predicted near-surface wind directions and wind stresses will have a large impact on predicted sand (and dust) fluxes, predicted sand transport directions, and hence predicted aeolian activity and features, including dust storms. While more comprehensive measure- ments are desirable, near-surface measurements of bulk variables at one height do provide a
crucial means of determining which models are performing best, from which we can try to understand why and learn how to improve models in general.
Combining aeolian and meteorological measurements will allow us to better understand the physics of aeolian processes in the Jezero region, and on Mars in general. Obtaining an estimate of the threshold wind stress required for saltation will help to explain currently active aeolian features all across Mars and the physical processes involved in triggering dust storms. It will also shed light on the potential past climates needed to explain the aeo- lian record, as preserved by the characteristics of depositional bedforms, erosional features, and some sedimentary rocks (Mars 2020 Objective A). Past saltation may have affected the distribution of micro-organisms in the near-surface, potentially by destroying radiation- or oxygen-resistant organisms via saltation-mediated abrasion (Bak et al.2019). Present day wind and aeolian measurements should enable a clearer understanding of the age of surface aeolian features, which may then be relevant to the search for materials with high biosig- nature preservation potential (Mars 2020 Objective B.2). Environmental measurements will also help to constrain the weathering and preservation potential of a possible cache sample (Mars 2020 Objective C.1). Linking the occurrence of strong wind stresses and/or convec- tive vortices to observed changes in dust cover will help us understand how dust is removed from the surface of Mars. Understanding dust lifting processes is a fundamental requirement for being able to correctly simulate – and perhaps one day predict – major dust storms. Be- cause dust is the largest driver of Martian weather and climate variability (e.g. Read et al.
2015; Forget and Montabone2017; Martínez et al.2017), this has great importance for predicting atmospheric conditions that may be encountered by future robotic and manned missions (Objective D), especially at critical times such as during Aerobraking and EDL.
In addition, better understanding of dust storms, dust lifting, sand motion, and the particle sizes and fluxes involved is important to both robotic and human surface operations, with the impact of dust on health especially important in the latter case.
In this work, we present predictions of the meteorology and aeolian activity at the Mars 2020 landing site in Jezero crater from nine different simulations performed using seven Mars atmospheric models. Four of the simulations are run at high spatial resolution, be- tween∼1.4 and 10 km horizontal grid spacing, while the remaining five are run at rela- tively low spatial resolution, between∼120 and 300 km. We focus on the diurnal cycles of pressure, temperature (air and surface), and wind (speed and direction) at the landing site at the time of landing, areocentric solar longitude Ls∼5◦, which corresponds to early northern spring, placing these results into context by examining the regional circulation. We also examine the expected seasonal variation in pressure, temperature, and circulation at the landing site, again placing them in regional context. We further provide aeolian predictions for the landing site from eight simulations, and for the wider Jezero crater region for the two highest-resolution simulations shown in this paper, mesoscale MarsWRF and MRAMS (see Table1), comparing with aeolian features observed from orbit. We highlight the dif- ferences in model predictions, explaining them where possible, and demonstrate the value of Mars 2020 data for increasing our understanding of and predictive skill for the Martian near-surface circulation and aeolian activity. A companion paper, Pla-García et al. (2020, this journal), examines the detailed meteorology expected at different seasons in more de- tail, including the expected water cycle, again using results from the two highest-resolution model simulations shown here.
Section2briefly describes the Mars 2020 instruments and measurements that will pro- vide meteorological and aeolian information that may be compared with the results of this study. Section3describes the seven atmospheric models and how they were set up to pro- duce the nine simulations used for this intercomparison. Section 4describes how mete- orological observations or model predictions may be combined with aeolian theory and
assumed surface properties to predict aeolian features and activity. Section5presents multi- model meteorological predictions for the Mars 2020 landing site at the landing date of Ls∼5◦, while Sect. 6 presents predictions for the landing site as a function of season.
Section7uses output from two high-resolution simulations to predict the changes in circu- lation that may occur if the rover drives out of Jezero crater toward the Midway and NE Syrtis locations. Section8predicts sand fluxes and bedform migration rates and orientations at the landing site and across the Jezero region. Section9predicts dust devil activity in the Jezero region and compares with other landing sites. Section10compares aeolian predic- tions over the Jezero region with orbital observations of aeolian features and their motion.
Finally, Sect.11summarizes results and concludes.
2 Instruments and Measurements Relevant to Meteorological and Aeolian Processes
In this section we provide an overview of the Mars 2020 sensors that will be used to measure the meteorological variables and aeolian activity predicted in this paper. In addition to the measurements described below, Mars 2020 instruments will also measure water vapor rela- tive humidity and column abundances, other trace gas abundances, and water ice properties.
Those measurements are not described here as this paper does not include predictions of those quantities. That does not mean such quantities cannot be predicted by this set of mod- els, rather that we chose to focus on the basic meteorological variables, circulation patterns, and aeolian activity. Indeed, five of the simulations (Ames x 2, GEM-Mars, OpenMARS, and global LMD; see Table1) predicted the seasonal and diurnal variation of water vapor, but such predictions remain very sensitive to assumptions and parameters included in the models, thus were deliberately excluded. However, Pla-García et al. (2020, this journal) examine the expected variation of water vapor in Jezero crater guided by orbital observa- tions, and a comparison of predicted water vapor abundance should form part of any future intercomparison study.
2.1 The Mars Environmental Dynamics Analyzer (MEDA)
The primary meteorological instrument on Mars 2020 is the Mars Environmental Dynamics Analyzer (MEDA). In several respects, MEDA is similar to MSL’s Rover Environmental Monitoring Station (REMS) (Gómez-Elvira et al.2014), but it has important additions and improvements (Rodriguez-Manfredi et al.2020, this journal). Unlike REMS, which had very limited memory, MEDA will be able to measure a complete daily meteorological cycle at 1 Hz (or for some sensors, 2 Hz) frequency. The planned MEDA baseline is to measure at least 12 hours per sol at a frequency of 1 Hz (Rodriguez-Manfredi et al.2020, this journal), although this may be limited by the power or data storage/downlink available.
2.2 Temperature and Pressure
The MEDA pressure sensor (PS) sits inside the rover body, connected to a tube and HEPA filter with a geometry that is designed to be insensitive to wind velocity. The MEDA atmo- spheric temperature sensors (ATS) are mounted at several heights on the rover, including just below the two wind sensor booms at∼1.5 m on the Remote Sensing Mast (RSM) and on the sides of the rover body, at 0.5 m. The MEDA Thermal InfraRed Sensor (TIRS) will also provide an estimate of atmospheric temperature over a region centered at∼40 m altitude, as well as measuring surface brightness temperature about 3 m from the rover.
Pressure and air temperature can be used to estimate air density using the ideal gas equa- tion, a vital requirement for estimating wind stress (see Sect.2.6). These variables, in ad- dition to wind speed and direction, also react strongly to the passage of clear or dusty con- vective vortices. Pressure data are particularly useful, due to the distinctive, rapid pressure drop and recovery as a vortex passes over or close to the sensor, hence pressure data have been commonly used to identify the statistics of vortex occurrence on Mars (e.g. Schofield et al.1997; Ellehøj et al.2010; Kahanpää et al.2016; Steakley and Murphy2016; Ordoñez- Exteberria et al.2018; Newman et al.2019a).
2.3 Radiative Fluxes and Aerosols
Between them, TIRS and the Radiation and Dust Sensor (RDS) also included in MEDA will constrain solar and IR upward and downward fluxes, surface properties, and aerosol (dust and to a lesser extent water ice) abundances and radiative properties at the rover’s location (Rodriguez-Manfredi et al.2020, this journal). In addition, the Mastcam-Z cam- eras mounted on the RSM (Bell et al.2020), which are able to provide color images and video in any direction with a powerful zoom capability, will measure aerosol (dust and wa- ter ice) abundance, vertical distribution, size distribution, and optical properties, as done using Mastcam on MSL (Lemmon et al.2019). The angular distribution of sky brightness observed by the Navigation (Navcam) and Hazard Avoidance (Hazcam) cameras may also be used to determine the aerosol abundances and properties, again as on MSL (Chen-Chen et al.2019a,b). Finally, the SuperCam instrument (Wiens et al.2020, this journal) will be used to obtain aerosol abundances and properties, as done using the ChemCam instrument on MSL (McConnochie et al.2018). In combination, this information on radiative fluxes and aerosol abundances and properties will aid interpretation of the meteorological mea- surements and enable future atmospheric modeling to be performed using more realistic local surface properties and aerosol forcing.
2.4 Wind
Like REMS, MEDA has two wind sensor booms mounted on the RSM at∼1.6 m altitude, which point at 120◦to each other in the horizontal plane (see Fig. 29 of Rodriguez-Manfredi et al.2020, this journal). This is required to measure winds correctly from all directions be- cause the RSM strongly perturbs wind that arrives at it before the sensor. Unfortunately, major damage to the wind sensors on the side-/rear-pointing REMS boom on landing pro- duced large gaps and biases in the wind dataset (Gómez-Elvira et al.2014; Newman et al.
2017; Viúdez-Moreiras et al.2019a,b). Further damage to the wind sensors on the front- pointing boom on MSL sol∼470 (∼2.2 Mars years into the MSL mission) meant that no REMS wind data have been available since September 2016. In addition, the front-pointing wind sensor boom suffered major electronic noise for temperatures below∼210 K, which meant that wind measurements could not be obtained for between 6 and 10 hours overnight, depending on season.
The design of the MEDA booms addresses many of the above issues. Where REMS has three wind sensor boards arranged around each boom, MEDA has six, providing increased redundancy. MEDA’s side-/rear-pointing boom is also longer than on REMS and unfolds post-landing, which both provides better damage protection and allows wind measurements to be made further away from flow interruptions caused by the RSM. In addition, as already done for InSight’s wind sensors (Velasco and Rodríguez-Manfredi2015), the design of the electronics has been improved, enabling calibration of the wind sensors for wind speeds of
up to 40 m/s (versus∼20 m/s for REMS) and reducing noise (Rodriguez-Manfredi et al.
2020, this journal). Mars 2020 also carries a microphone on SuperCAM that may be able to determine wind speed from the intensity of the noise at low frequencies (<500 Hz) and wind direction from the relative intensity of the noise measured at different rover pointings (Chide et al.2021). These measurements will be used for cross-calibration with MEDA’s wind sensors where possible. Finally, the Ingenuity Helicopter’s flight control sensors include measurements of air pressure and vehicle acceleration that should provide insight into the winds that it encounters at various levels during flight. This may provide an idea of the vertical wind profile near the surface of Mars, and hence an alternative estimate of wind stress using the flux profile method described in Sect.2.6.
Convective vortices (either clear or dusty) are formed of rotating air around a central, low-pressure vortex core. They therefore produce a perturbation, in both wind speed and direction, that changes sign as a vortex passes over the wind sensor. This signal may be harder to detect when the wind is highly turbulent, or when a vortex does not pass directly over the sensor, but Kahanpää et al. (2016) noted that 87% of vortices detected via their pressure drops also had a strong signature in wind direction. Note that, while Coriolis forces constrain much larger vortex features (e.g. hurricanes) to rotate either clockwise or anti- clockwise depending on the hemisphere, for dust devil-sized vortices no preferred sense of rotation has generally been observed and only a tiny impact on sense of rotation is predicted by Large Eddy Simulation modeling (Ito et al. 2011). However, in situations where the mesoscale circulation is cyclonic or anticyclonic due to local weather patterns, dust devil- sized vortices are predicted – and have been observed – to share the same sense of rotation (Ito et al.2011; Fujiwara et al.2012).
2.5 Aeolian Features and Activity
As on all past landed missions, the Mars 2020 rover does not carry any dedicated aeolian instruments, such as saltation detectors or sand/dust flux sensors. However, images from Mars 2020 cameras may be used to study the properties of sand grains and aeolian fea- tures, their motion, surface albedo changes, and dust devils or other dust-raising activity.
High-frequency meteorological data may then be used to infer the cause of observed aeolian activity.
2.5.1 Orientation and Motion of Surface Aeolian Features
Like Mastcam, its predecessor on MSL, Mastcam-Z will be used to obtain high-resolution views of surface aeolian features (e.g. ripples, ventifacts, and dunes). In addition to de- termining the characteristics of these features, such images will also be used for “change detection” experiments, in which images taken of the same surface region at different times are co-registered to look for changes, which are then attributed to aeolian processes. These techniques have been used to determine the seasonal and diurnal variation in aeolian activ- ity in Gale crater, especially within the Bagnold Dune Field (Bridges and Ehlmann2017;
Bridges et al.2017; Baker et al.2018a,b). Like MSL’s Curiosity rover, Mars 2020’s Perse- verance rover also carries two Navcams and six Hazcams, which are enhanced by having a wider field of view and imaging in color. While these cameras have a lower resolution than Mastcam-Z, they are far less in demand when the rover isn’t driving, hence are good choices for making more frequent aeolian observations (Greeley et al.2010; Baker et al.2018b).
Finally, Mars 2020 will also carry a suite of Descent Imagers, one of which will look down at the surface from underneath the rover, much like the Mars Descent Imager (MARDI) on
MSL. While this camera is intended primarily for EDL, it may also be used after landing to image the surface beneath the rover when it is not in shadow, specifically around sunrise and sunset, extending imaging further into the early morning and late evening to help iden- tify the timing of observed changes (Baker et al.2018b). The size distribution of particles involved in aeolian activity is vital for interpreting such changes (Baker et al.2018b; Weitz et al.2018). On Mars 2020, detailed images of surface grains down to tens of microns will be available from the Scanning Habitable Environments with Raman and Luminescence for Organics and Chemicals (SHERLOC) micro-imagers at the end of the robotic arm (Bhartia et al.2020, this journal).
Aeolian investigations on MSL have been hampered by a lack of good wind data through- out the mission (see Sect.2.4). InSight has benefited from improved wind data, but has lower resolution cameras and lacks any significant aeolian features at its landing site (Golombek et al.2020). Despite this, albedo and other changes seen by InSight have been used to ex- plore possible thresholds for particle motion and to attempt to differentiate between large- scale wind-induced and vortex-induced lifting (Baker et al.2020; Charalambous et al.2021).
This suggests that a combination of good wind (and pressure) data, higher-quality imaging, and more access to aeolian features for the Mars 2020 mission could provide enormous in- sight into aeolian processes and the relative importance of “wind stress” and “dust devil”
dust lifting (Newman et al.2002a,b). On longer timescales, a wind stress dataset spanning a full Mars year can be used to predict the long-term direction of motion of local dunes and the orientation of local aeolian features, including dunes, ventifacts, and yardangs. This can be used to understand both the characteristics of currently active features and to indicate that other features most likely formed under past climate conditions.
2.5.2 Dust Devils (Dusty Convective Vortices)
Multiple images of the same view have been used on all Mars surface missions since Mars Pathfinder to detect dust devils (Metzger et al.1999; Greeley et al.2010; Lemmon et al.
2017). Similar to MSL, the majority of dust devil monitoring on Mars 2020 will be per- formed using the Navcams, Hazcams and Mastcam-Z. This monitoring, which will consist of regular surveys (a few images covering all directions) and movies/videos (multiple images or a Mastcam-Z video covering up to 30 minutes in one direction) has two purposes: (i) to study the statistics of how dust devil number and size varies with time of sol, season, and location, and (ii) to determine the characteristics of dust devils, such as their dust content, direction of motion (indicating winds at that location), and height (which may be related to the height of the PBL; Fenton and Lorenz2015). Whenever possible, simultaneous meteoro- logical data are taken, so that if the dust devil vortex passes over (or near to) the rover then its impact on pressure, wind, temperature, and radiative fluxes can be correlated with the imag- ing and used to infer more about the dust devil’s characteristics. These measurements are important for relating vortex activity to other atmospheric variables (Newman et al.2019a;
Spiga et al.2020); for understanding when and where surface dust is lifted, especially out- side of dust storms when dust devil lifting may dominate (Basu et al.2004; Kahre et al.
2006); and for understanding the likelihood of dust-devil-induced cleaning events on Mars 2020 and other missions, especially those that rely on cleaning of dust from solar panels to maintain power.
2.6 Obtaining Wind Stress from Mars 2020 Meteorological Data Wind stress,τ, is the primary driver of aeolian activity, and is given by:
τ=ρu2∗ (1)
whereu∗ is drag (or friction) velocity, andρ is air density, which is given byP /(RT ), whereP is surface pressure,R=191.272 J kg−1K−1is the specific gas constant for Mars, and T is near-surface air temperature. Note that u∗ can be directly measured from very accurate, high frequency measurements of the 3-D wind field near the surface, via equation:
u∗= uw2
+
vw21/4
(2) whereu,v, andw are the turbulent fluctuations of the three wind components and the overbar represents Reynolds (i.e. time) averaging. The highest frequency range would be achievable with the accuracy and frequency of e.g. a sonic anemometer, which can provide faster sampling than MEDA’s wind sensors. However, MEDA’s 1 Hz is sufficient to reach into the inertial range given the large Kolmogorov scale on Mars (e.g. Tillman et al.1994;
Schofield et al.1997; Murdoch et al.2017).
The alternative is to estimateu∗using a Businger-Dyer similarity relationship between the surface momentum flux and the mean vertical profile (Businger et al.1971; Stull1988).
Specifically,u∗ is related to the measured wind, u, at some height, z, and the estimated surface roughness,z0, by:
u(z)=u∗ κ
ln
z−d z0
+ψ (z, z0, L) (3)
whereκ is the Von Kármán constant (taken to be 0.4),d is the zero plane displacement (the height in meters above the ground at which zero wind speed is achieved as a result of flow obstacles),Lis the Obukhov length from Monin-Obukhov similarity theory, andψis a stability term. In the absence of obstacles, d may be taken to be zero. Another simplifying assumption is to assume neutral stability, in which case ψ is also 0. This assumption is likely to be incorrect for much of the Martian sol, e.g. during periods of convection (when the atmosphere is unstable) or at night if a strong inversion develops (hence the atmosphere is stable). However, under these assumptions, Eq. (3) becomes:
u(z)=u∗ κ
ln
z z0
≡u∗
κ ln(z)−u∗
κ ln(z0) (4)
which requires onlyz0to be estimated.
Estimatingz0 – especially for another planetary body – is not trivial, however. A vis- cous sublayer can exist near the surface in which surface friction causes viscous forces to dominate over inertial forces, resulting in smooth flow. The roughness Reynolds number, Rer=ρksu∗/μ, dictates the extent to which roughness elements on the surface disrupt this flow and its value determines howz0should be estimated. Hereμis the dynamic viscosity andksis the Nikuradse roughness (Nikuradse1933; White2006), which is approximately equal to particle diameter,Dp, for a homogeneous bed of monodisperse spherical particles, but more generally given by two to five times the median particle size. For Rer>60, the flow is “aerodynamically rough” (i.e.,Dpis large enough that turbulent mixing destroys the viscous sublayer) andz0≈ks/30. For Rer<4, the flow is “aerodynamically smooth” andz0
is given by the thickness of the viscous sublayer, which depends onu∗and is generally much larger. Due to its much lower atmospheric density than Earth, Mars is likely in the “smooth”
regime over surfaces with small roughness elements (e.g. a smooth bed of sand). However, estimates ofz0for Mars typically use the “rough” regime definition; see Kok et al. (2012) for discussion of why this is likely appropriate when saltation occurs. Surface roughness maps are therefore produced by assuming thatz0≈ks/30 and using orbital datasets related
to the height and spacing of roughness elements to estimateks. Methods include using Mars Orbiter Laser Altimeter (MOLA) topography and roughness maps (Heavens et al.2008) and Mars Global Surveyor (MGS) Thermal Emission Spectrometer (TES) rock abundance maps (Hébrard et al.2012).
A further concern is that the functional forms of the stability terms,ψ, and the estima- tions used forz0are empirically derived for Earth and are not yet known to work the same way under Martian conditions. Even assuming neutral conditions, for which Eq. (4) may be used and stability terms neglected, the uncertainty inz0is a concern. A more accurate esti- mate ofu∗– referred to as the flux profile method (e.g. Bi et al.2015) – is useful as it also provides an estimate ofz0under the neutral stability assumption. The method involves mea- suringu(z)at several heights, from whichu∗/κis obtained as the slope of the least squares fitting of uand ln(z); see RHS of Eq. (4). The intercept on the y-axis is then uκ∗ln(z0).
Unfortunately, given that MEDA measures wind at a fixed height, this method cannot be applied on Mars 2020, but it was applied to Mars Pathfinder wind sock data, which were available at three heights (Sullivan et al.2000).
Finally, it is important to note that, unlike Earth, the air density on Mars may change by several tens of percent from day to night, and by an even greater amount with season, due to the strong diurnal and seasonal variations in surface pressure and temperature. Hence the force exerted on surface particles, as given by the wind stress, is not simply related to wind speed or evenu∗. This means that wind speed is not a completely reliable indicator of when saltation is expected to occur. For example, it is possible for the peak wind stress to occur overnight (when densities are highest due to low air temperatures) while peak wind speeds occur during the daytime (e.g. Baker et al.2018a,2021). Thus it is critical to consider the estimated wind stress, rather than simply the wind speed at∼1.6 m, when interpreting aeolian features and activity.
3 Atmospheric Models and Setups Used in This Study
This study uses output from seven different Mars atmospheric models, two of which (Ames and MarsWRF, Sects.3.4and3.5) are run in both “low-resolution” and “high-resolution”
mode. This gives a total of nine simulations, with four run at high spatial resolution over Jezero crater (between∼1.4 and 10 km horizontal grid spacing) and the others run at rel- atively low spatial resolution (between∼120 and 300 km); see Table1. Due to the high computational expense of running at high resolution, the full seasonal cycle is only predicted using the low-resolution simulations, while it is sampled at the landing time (∼northern spring equinox) and at up to eleven other times of year by the high-resolution simulations.
Sections 3.1–3.7describe the seven models and provides more details of how they were set up for this intercomparison study, while Sect.3.8describes how the model output was processed to provide a direct comparison with the fields that Mars 2020 will observe.
Unless otherwise specified, all models used surface topography derived from MOLA data (Smith and Zuber1996; Smith et al.2001), while surface albedo and thermal inertia over most of the surface are derived from MGS TES observations (Christensen et al.2001;
Putzig et al. 2005; Putzig and Mellon 2007), although polar ice properties are typically used as tuning parameters for the CO2cycle (see Sect.6.1). Despite this, the surface albedo and thermal inertia interpolated to the landing site differ between models. This is due to how the observed values are interpolated and smoothed onto the model grid and due to the size of the model grid cell in each simulation. Aerodynamic surface roughness,z0, is set to be spatially uniform or according to a map derived from MOLA intrashot data (Garvin
Table1Horizontalgridspacing,surfacepropertiesatthelandingsite,heightofthemidpointofthelowestmodellayer,andheighttowhichatmospherictemperature(Tair)is extrapolated,fortheninesimulationspresentedhere.SeeaboveandSects.3.1–3.7formoreexplanation SimulationHorizontalgrid spacingAlbedoThermalinertia (Jm−2K−1s−1/2)EmissivityRoughness map/methodRoughnessheight, z0(m)Heightoflowest modellayer(m)Tairgiven at(m) GEM-Mars4◦(237km)0.1935288.50.950Hébrard0.007422 LMD(global)5◦(296km)0.186283.00.950Hébrard0.00784.54.5 OpenMARS5◦(296km)0.1667321.00.950Hébrard0.011051.5 Ames(lowresolution)1.875◦(111km)0.1540284.40.944Uniform0.010055 MarsWRF(global)2◦(118km)0.1342260.70.950Garvin0.0267101.5 LMD(mesoscale)10km0.165319.00.950Hébrard0.01104.54.5 Ames(highresolution)6.6km0.1721293.40.941Uniform0.010055 MarsWRF(domain5)1.4km0.1342260.70.950Garvin0.0267101.5 MRAMS2.96km0.1362260.00.950Garvin0.030014.514.5
et al.1999) or MGS TES rock abundance data (Hébrard et al.2012), as shown in Table1.
Table1also provides the grid spacing and surface properties at the landing site in all nine simulations and is a useful reference for interpreting the differences between model results described in Sects.5and6. While surface properties over a wider area will also influence the circulation observed at the landing site, local values of albedo and thermal inertia will have the largest impact on surface and atmospheric temperature, and – as on past missions – such observations will be used to infer the variation of these surface properties along the rover traverse (Hamilton et al.2014; Vasavada et al.2017).
Atmospheric dust content is a major control on the atmospheric thermal state and thus circulation of Mars (e.g. Gierasch and Goody1972; Wilson and Hamilton1996; Kahre et al.
2017; Wolff et al.2017). Dust content can vary significantly from year to year, making it by far the main source of interannual variability in Martian climate. The greatest year-to-year variations occur primarily during the so-called “dust storm season,” Ls∼180 to 360◦. Dust loading affects the strength of the large-scale circulation and thermal state, and changes in the horizontal and/or vertical variation of dust affect thermal tides and other planetary-scale waves, modifying diurnal cycles of pressure and winds at the surface. For these reasons, all simulations presented here were conducted using a seasonal variation of dust loading that corresponded to no major dust storms occurring in that year. However, the specific dust distributions were left to each modeling group to decide and ultimately reflect a broad set of choices for what best represents a “storm-free” year. The dust distributions and their evolution with time are described in each of the following sections. In addition, Table 2 provides the visible column dust opacity over the landing site at four times of year, but note that the vertical dust distribution and particle size distribution may also be crucial.
While specifying the dust identically in all model simulations would have permitted the most direct assessment of the impact of model dynamics, physics, and resolution on results, this was not possible given the scope and timeline of this study. However, the simulations shown here are those that each modeling group commonly use as their “storm-free” predic- tions, hence any additional spread in results is representative of the spread in model predic- tions that have been and may in future be used for a variety of scientific and engineering purposes. In addition, this enables us to compare the impact of different approaches used to specify dust.
3.1 The GEM-Mars Atmospheric Model
The GEM-Mars model (Neary and Daerden2018; Daerden et al.2019; Neary et al.2020) is a gridpoint-based general circulation model of the Mars atmosphere based on the GEM (Global Environmental Multiscale) model, part of the operational weather forecasting and data assimilation system for Canada. The model extends from the surface to approximately 150 km and simulates interactive carbon dioxide, dust, water, and atmospheric chemistry cycles. Dust and water ice clouds are radiatively active. The dynamical core uses a semi- Lagrangian advection scheme with a two-time-level semi-implicit integration method that allows for a relatively long time step while maintaining stability. The simulations for this study were performed at a horizontal resolution of 4◦×4◦with 103 unevenly-spaced log- hydrostatic pressure levels. The height of the lowest atmospheric layer is set to 2 m, with the next level up at∼13 m. A time step of 1/48th of a Mars solar day (sol) was used.
3.1.1 Dust Distribution in the GEM-Mars Simulation
This simulation is the only one in the study to have a 3-D dust distribution that is fully self- consistent with the model circulation, rather than being directly constrained by observations.
Table 2 Daily mean visible column dust opacity over the landing site at the four key seasons examined in Sects.5and6, and a summary of the dust scenario used, for each of the nine simulations
Simulation Visible dust opacity for season shown Dust scenario used (see Sects.3.1.1 to3.7.1for more details)
Ls∼5◦ Ls∼90◦ Ls∼180◦ Ls∼270◦
GEM-Mars 0.26 0.18 0.19 0.52 Interactive dust with dust devil and
saltation lifting
LMD global 0.31 0.26 0.35 0.62 Interactive dust but rescaled to
match dust maps from years without global storms
OpenMARS 0.32 0.26 0.30 0.61 Assimilated temperature and dust
from MY32
Ames low-resolution 0.30 0.26 0.4 0.63 Interactive dust but with lifting constrained to match MY30 dust opacity map
MarsWRF global 0.30 0.19 0.33 0.44 Prescribed using TES nadir and limb dust opacities for years without global storms
LMD mesoscale 0.31 N/A N/A N/A As in LMD global model
Ames high-resolution 0.30 0.26 0.4 0.63 As in Ames low-resolution model MarsWRF mesoscale 0.30 0.19 0.33 0.44 As in MarsWRF global model
MRAMS 0.33 0.22 0.23 0.54 Prescribed using MY24 TES nadir
dust maps and SPICAM vertical profiles
Size-distributed dust is injected according to parameterized saltation and dust devil dust lifting, followed by dust transport by advection, mixing, and sedimentation, as in (Musiolik et al.2018). The saltation dust lifting parameterization uses a lower threshold wind stress than is typical, based on results of low-gravity experiments, and the overall dust lifting parameters are then tuned such that running the simulation provides a generally realistic variation of dust loading over the year. There is a gradual increase in dust loading after about southern spring equinox, especially in the southern hemisphere, peaking at a visible (0.67 micron) opacity of∼0.8 at Ls∼270◦at∼40◦S latitude (Musiolik et al.2018).
3.2 The LMD Global Atmospheric Model
The Laboratoire de Météorologie Dynamique (LMD) Mars global circulation model (GCM) has been developed over the past 25 years (Forget et al.1999; Lewis et al.1999) in collab- oration with LMD, Oxford University, the Open University (OU), and the Instituto de As- trofisica de Andalucía. Fluid dynamics equations for the atmosphere are solved in a finite- difference grid point dynamical core. Physical parameterizations of processes unresolved by the hydrodynamical solver include a representation of the most salient characteristics of the carbon dioxide, dust, and water cycles, with dust and water being transported by the model circulation and with both dust and water-ice particles being radiatively active (Madeleine et al.2011; Navarro et al.2014). A two-moment scheme is used for dust particles; this means that the dust mixing ratio and number concentration are both carried, such that the dust particle size distribution at any time and location may be inferred. The GCM includes the radiative effects of water ice clouds (Madeleine et al.2012) that are produced by a com- plete water cycle model using a microphysical scheme that calculates the growth of water
ice crystals onto dust nucleation cores (Navarro et al.2014). The PBL daytime convective processes are represented by a specific “thermal plume” model described in Colaïtis et al.
(2013). The baseline simulation has 64 by 48 grid points, corresponding to a resolution of 5.625◦in longitude by 3.75◦in latitude. The vertical layers are distributed using a hybrid sigma-pressure coordinate system with the first model layer centered at∼4.5 m; the model top is located at∼250 km.
3.2.1 Dust Distribution in the LMD GCM Simulation
In this simulation, the horizontal distribution is constrained to match observations from years without major dust storms, with the vertical distribution determined more self-consistently by the model. In the LMD model’s semi-interactive dust scheme, two-moment dust lifting is performed according to dust devil and wind stress parameterizations, but the atmospheric dust column is then rescaled to match daily maps obtained from observations by orbiting spacecraft (Madeleine et al.2011). For this simulation, column dust opacities averaged over MYs 24, 26, 27, 29, 30 and 31 (well-observed years with no major storms) were used to produce average “storm-free year” dust maps as a function of season (Montabone et al.
2015).
3.3 The OpenMARS Database
Data were taken from GCM assimilations of previous Mars years, as archived in the OU’s OpenMARS database (Holmes et al.2019,2020). Data assimilation is conducted for col- umn dust opacities and thermal profiles (Lewis et al.2007; Montabone et al.2006), with water vapor and ice (Steele et al.2014a,b), ozone (Holmes et al.2018), and carbon monox- ide (Holmes et al.2019) also assimilated when available. The model used for assimilation is the OU version of the LMD GCM, which shares the LMD GCM’s surface property maps and physical sub-models (see Sect.3.2) but uses a semi-spectral dynamical core and a semi- Lagrangian conservative tracer transport scheme (Newman et al.2002a,b). It also includes a gravity wave drag scheme that includes low-level drag from sub-gridscale orography de- rived from the MOLA 1/32◦data set (Collins et al.1997). Although the model has a full water cycle scheme, and TES water data were assimilated, the simulation did not include radiatively-active ice clouds, in order to ensure stability. The assimilations used were con- ducted using a spectral truncation of dynamical model fields at wavenumber 31 (T31, with a 3.75◦horizontal grid for dynamical products) and 5◦(300 km) horizontal grid for the phys- ical sub-models, with 35 vertical sigma levels stretched up to about 100 km altitude and the lowest level at∼5 m above the surface.
3.3.1 Dust Distribution in the OpenMARS Simulation
As for the LMD GCM simulation, here the horizontal distribution is constrained to match observations from years without major dust storms, with the vertical distribution determined more self-consistently by the model. Temperature profiles and total column dust opacities measured by MCS in MY32, a year with no major dust storm, are assimilated. The verti- cal dust distribution is determined via the semi-interactive scheme described in Sect.3.2.1 (Madeleine et al.2011), with the atmospheric dust column at each grid point (i.e. the hor- izontal dust distribution) now rescaled to match MY32 observations via the assimilation process.
